AI callers leave contact centers scrambling, TTEC warns
LANCASHIRE, UNITED KINGDOM — Contact centers are struggling to handle a new type of caller: AI agents acting on behalf of real customers. At a major United States bank, an AI completed a debt negotiation without human involvement, leaving staff unsure how to respond.
According to a report from CX Today, the situation demonstrates that contact centers face a mounting challenge as their existing systems and workflows lack the capability to handle AI-based interactions.
AI callers expose cracks in traditional contact centers
Wayne Kay, Regional Vice President of Sales Leadership EMEA at TTEC Digital, said the realization is spreading across industries.
“The agent didn’t know what to do. There was no policy internally for handling it. They didn’t hang up. They went through authentication and concluded a debt negotiation with the AI agent. Then they hung up and said, ‘Oh my goodness, what do I do? How does this get followed up?’” he told the CCMA Tech Summit.
The problem stems from systems built exclusively for human interaction. Knowledge-based authentication fails against AI, which recalls information perfectly, while fraud detection tools can trigger false positives.
Interactive voice response systems, designed to interpret human speech patterns, may misread AI precision as scripted fraud. Even analytics and performance metrics struggle, as it becomes difficult to distinguish humans from AI interactions.
3 steps to build an AI-ready strategy
To address these challenges, Kay outlined a three-step framework for AI readiness. Authentication comes first, moving beyond passwords to token-based, cryptographic protocols.
“Authentication comes first. Is this real? Do they have permission?” he said.
Knowledge management follows, ensuring AI agents only access information appropriate for their tasks.
“You look at your knowledge that you give them access to. Is it minimum permissible rights? Can this AI only get access to what it should have access to?” Kay explained.
Operational redesign is the final pillar, creating dual workflows: transactional lanes for AI requiring speed and accuracy, and empathetic lanes for human customers needing reassurance.
Organizations are also advised to start with policy and sandbox testing before full implementation.
“Treat every inbound call as potentially coming from an automated autonomous agent, not a human,” Kay emphasized.
The outsourcing industry needs to adapt as AI systems now serve as real customer presence. Contact centers that operate on international client bases require advanced AI-ready systems which combine secure and efficient AI-human interaction methods with human-centered empathy solutions.
Early adopters who redesign authentication, knowledge access, and operational workflows are likely to gain a competitive edge, while those relying solely on human-centric processes risk inefficiency and security exposure.
